Scale disambiguation of blindly unmixed endmember abundances

Salerno E.

Noisy linear mixture models  Scale ambiguity  Spectral unmixing  Remote sensing 

The problem of estimating the scale factors related to blindly separated relative abundance maps is addressed. The nature of the quantities looked for is such that their sum must be everywhere equal to one. Exploiting this property, the problem is solved within the framework of a linear noisy mixture model, with signal-independent and zero-mean noise.

Source: ISTI Technical reports, 2006